Transcriptional activity of interferon regulatory factor (IRF)-3 depends on multiple protein–protein interactions Hongmei Yang 1 , Charles H. Lin 2 , Gang Ma 1 , Melissa Orr 1 , Michael O. Baffi 1 and Marc G. Wathelet 1 1 Department of Molecular and Cellular Physiology, University of Cincinnati College of Medicine, Cincinnati; 2 Department of Molecular and Cellular Biology, Harvard University, Cambridge, MA Virus infection results in the activation of a set of cellular genes involved in host antiviral defense. IRF-3 has been identified as a critical transcription factor in this process. The activation mechanism of IRF-3 is not fully elucidated, yet it involves a conformational change triggered by the virus- dependent phosphorylation of its C-terminus. This con- formational change leads to nuclear accumulation, DNA binding and transcriptional transactivation. Here we show that two distinct sets of Ser/Thr residues of IRF-3, on phosphorylation, synergize functionally to achieve maximal activation. Remarkably, we find that activated IRF-3 lacks transcriptional activity, but activates transcription entirely through the recruitment of the p300/CBP coactivators. Moreover, we show that two separate domains of IRF-3 interact with several distinct regions of p300/CBP. Interfer- ence with any of these interactions leads to a complete loss of transcriptional activity, suggesting that a bivalent interaction is essential for coactivator recruitment by IRF-3. Keywords: interferon; IRF-3; coactivator; virus; transcrip- tion. Vertebrates respond to infections by first triggering the innate arm of the immune system upon recognizing generic microbial products, such as lipopolysaccharides for Gram- negative bacteria, or dsRNA for viruses [1,2]. A number of cytokines, including interferons (IFNs), tumor necrosis factors, and several chemokines and interleukins, are produced early upon infection. They signal to the organism the presence of the infection, and alter the behaviour of a large number of cells in order to expedite pathogen elimination. The production and action of these cytokines depends in large part on specific modulations of gene expression. IFN regulatory factors (IRFs) have extensive roles in innate immunity by participating in both the immediate- early transcriptional response to infection and the secondary response to cytokines [3–5]. For instance, IFN-b synthesis is activated directly by virus infection in most cell types, or by lipopolysaccharide in some specialized cells, processes requiring IRF-3 and IRF-7 [6,7]. The transcriptional response as activated by IFN-b in turn depends on IRF-1 and IRF-9 via the JAK-STAT pathway. IRF-3 is a latent transcriptional activator protein that becomes active only after being exposed to pertinent stimuli, as is the case for IRF-5 and IRF-7. By contrast, the activity of IRF-1 and IRF-9 is constitutive. The mechanism by which these virus-dependent IRFs are activated remains to be characterized, but it involves the phosphorylation of a stretch of serine (Ser) and threonine (Thr) residues at their C-terminal ends. This phosphorylation results in a con- formational change that allows nuclear accumulation, DNA-binding and transcriptional activation of target genes [8–17]. However, the identity of the functionally important phosphorylation targets remains controversial (Fig. 1A). Indeed, while Fujita and colleagues propose Ser385/386 to be the key residues in the activation of human (h)IRF-3 [16,18], Hiscott and colleagues point to the five Ser/Thr residues between amino acids 396–405 as the responsible targets [9,19]. Similarly, the identity of phosphorylated residues is unclear for IRF-7 [10,20]. The candidate residues in IRF-3 and IRF-7 fall into two groups: the first group comprises two Ser residues that are conserved among the human, murine and chicken proteins, while the second group comprises five or six Ser/Thr residues in a less conserved region immediately downstream from the first set (Fig. 1A). In addition, the nature and functional import- ance of protein–protein interactions in IRF-3-dependent transcriptional activity remains poorly defined. Here we first show that IRF-3 activation depends on synergy between the two sets of Ser/Thr residues. Modifi- cation of residues within both sets is required to achieve full DNA binding and transactivation capabilities. Intriguingly, we found that IRF-3 lacks intrinsic transcriptional activity, but activates transcription entirely through the recruitment of the p300/CREB-binding protein (CBP) coactivators. This was demonstrated using cells (insect Schneider S2 cells) that are devoid of endogenous IRF and where the endogenous Drosophila CBP sequence is functionally unable to Correspondence to M. G. Wathelet, Department of Molecular and Cellular Physiology, University of Cincinnati College of Medicine, 231 Albert Sabin Way, Cincinnati, OH45267-0576, Fax: + 513 558 5738, Tel.: + 513 558 4515, E-mail: marc.wathelet@uc.edu Abbreviations: CAT, chloramphenicol acetyl transferase; CREB, cAMP response element binding protein; CBP, CREB-binding pro- tein; EMSA, electrophoretic mobility shift assay; IFN, interferon; GST, glutathione-S-transferase; IRF, IFN regulatory factor; ISGF-3, IFN-stimulated gene factor 3; ISRE, IFN stimulated response element; STAT, signal transducer and activator of transcription; WT, wild-type. (Received 8 August 2002, revised 15 October 2002, accepted 23 October 2002) Eur. J. Biochem. 269, 6142–6151 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03330.x substitute for mammalian p300/CBP. Moreover, we found that IRF-3 interacts with several distinct regions of p300/ CBP. Finally, we show that these protein–protein interac- tions are essential for acquiring transcriptional activity. EXPERIMENTAL PROCEDURES Plasmid constructs and sequence analysis Effector constructs for transient transfections of mamma- lian and insect cells were cloned into pcDbA and pPac vectors, respectively, using standard methods [21]. In these constructs, the coding sequence of IRF-3 was preceded by a histidine tag, containing a stretch of six His residues (H6- IRF-3). Alternatively, the coding sequence of IRF-3 was fused to the Gal4 DNA-binding domain. Mutants of IRF-3 were generated by PCR and were all verified by sequencing. Reporter constructs have been described [14]. Cell culture and transfections HEC-1B (HTB-113, ATCC) cells are derived from a human endometrial carcinoma and are resistant to IFN; SAN cells are derived from a human glioblastoma and are lacking type I IFN genes; 293T cells are a SV40 large T antigen- expressing highly transfectable derivative of 293 cells, which are derived from human embryonic kidney cells trans- formed with human adenovirus type 5. These cell lines were grown at 37 °C, 5% CO 2 , in Dulbecco’s modified Eagle medium containing 10% fetal bovine serum, 50 UÆmL )1 penicillin and 50 lgÆmL )1 streptomycin. Sendai virus was obtained from SPAFAS and used at 200 hemaglutinin UÆmL )1 . S2 cells were grown at 26 °C, in Schneider’s Drosophila medium containing 12% fetal bovine serum, 50 UÆmL )1 penicillin and 50 lgÆmL )1 streptomycin. Transfections using the calcium phosphate coprecipita- tion technique were as described [21]. Mammalian cells in 100 mm dishes were transfected with 1 mL of a precipitate containing 10 lg reporter, 5 lg effector plasmid (except for Gal4, 1 lg), 5 lg pCMV-lacZ and pSP72 to a total of 25 lg (5–9 lg) for 18 h, trypsinized, aliquoted for further treat- ments and harvested 3 days after transfection. S2 cells were seeded in 6-well plates (3 million cells in 3 mL), transfected the next day with 0.3 mL of a precipitate containing 250 ng hsp82lacZ, 500 ng reporter plasmid and effector plasmid mixes as indicated in figure legends (with pPac added to a total of 5.75 lg), and harvested 2 days after transfection. Chloramphenicol acetyl transferase (CAT) activity and b-galactosidase activity were measured in extracts of transfected cells [21], and CAT activity was expressed in arbitrary units after normalization to b- galactosidase activity to control for transfection efficiency. In vitro translation, cell extract, EMSA and Western blot In vitro translation in rabbit reticulocyte lysates or wheat germ extracts was performed exactly according to the manufacturer’s recommendations using the TnT kit (Promega), linearized pcDbA effector plasmids and T7 RNA polymerase. Whole cell extract preparation, binding and PAGE conditions for electrophoretic mobility shift assays (EMSA) were as described [14], except that 5 pmoles of cold 9–27 IFN-stimulated response element (ISRE) were added for EMSA involving in vitro translated IRF-3. Immunoblotting, after SDS/PAGE or native gel electro- phoresis in the presence of deoxycholate [22], was performed as described [23], using mouse monoclonal SL-12 [anti- (IRF-3)] and rabbit polyclonal sc-510 (anti-Gal4, Santa Cruz) as primary antibodies, and anti-mouse or anti-rabbit horse radish peroxidase conjugates as secondary antibodies. The chemiluminescence detection system was from Perkin- Elmer life sciences. Pull-down experiments Glutathione S-transferase (GST)-CBP-N, -M, -C, -p300-N, -M and -C were described previously [24], GST-CBP-N1, -N2, -N3, -C1, -C2, -C3 and GST-p300-C1, -C2, -C3 were generated by subcloning PCR products and verified by sequencing. GST and GST fusions were expressed in E. coli BL21 and purified as recommended by the manufacturer (Pharmacia), and dialyzed against phosphate-buffered saline-10% glycerol. 35 S-Labeled in vitro translated proteins were incubated with GST fusion proteins immobilized on glutathione-sepharose beads in 150 m M KCl, 20 m M Tris pH 8.0, 0.5 m M dithiothreitol, 50 lgÆmL )1 ethidium bro- mide, 0.2% NP-40 and 0.2% BSA (binding buffer) for 1 h at 4 °C, followed by two washes with binding buffer and two washes with binding buffer without BSA. Extracts from HEC-1B cells labeled in vivo with [ 32 P]orthophosphate were prepared and treated with deoxycholate and NP-40 exactly as described [14]. After 50-fold dilution with binding buffer Fig. 1. Two sets of residues in IRF-3 are modified in response to virus infection. (A) Alignment of the C-terminal cluster of Ser/Thr residues that are potentially phosphorylated in virus-infected cells for human IRF-7B, murine IRF-7, human IRF-3, murine IRF-3 and chicken IRF-3. (B) Transcriptional activity in SAN cells of IRF-3 point mutants on the P31 ·2 CAT reporter (left graph) or as Gal4 fusions on the G5E1bCAT reporter (right graph); the sequence of WT and each point mutants is indicated on the left of the graphs; C: control, uninfected cells, V: virus-infected cells. Ó FEBS 2002 Mechanism of IRF-3 virus-dependent activation (Eur. J. Biochem. 269) 6143 and preclearing on glutathione-sepharose beads, the 32 P- labeled proteins were incubated with immobilized GST fusions for 2 h at 4 °C, followed by three washes with binding buffer. Proteins bound to the beads were eluted with radioimmunoprecipitation assay buffer and immuno- precipitated with SL-12 [anti-(IRF-3)] as described [14]. Pulled-down and immunoprecipitated proteins were then analyzed by SDS/PAGE and autoradiography. RESULTS Identifying the residues functionally involved in IRF-3 activation is important both for the characterization of the kinase(s) involved in activation and for our understanding of the activation mechanism. We undertook a systematic analysis of the role played by the two sets of Ser/Thr residues in the virus-dependent activation of IRF-3. To this end, we examined the phenotypes of a variety of hIRF-3 mutants by performing cotransfections in SAN cells. These cells are derived from a human glioblastoma and lack all type I IFN genes. The absence of IFN genes allows us to avoid the complication of a feed-back loop where virus- induced IFN in turn activates ISGF-3 (i.e. IFN-stimulated gene factor 3, a complex of IRF-9, STAT-1 and STAT-2) and STAT-1 dimers, leading to increased levels of endo- genous IRF-7 and IRF-1. Increased levels of IRF-1, IRF-7 and ISGF-3 would interfere with the activity of the reporter plasmid used in these experiments. We used the P31 ·2 CAT reporter, which is driven by the IRF-dependent element of the IFN-b gene promoter. This reporter is only weakly inducible by virus alone because of the relatively low affinity of IRF-3/7 for the P31 sequence (approximately 100-fold less than for an optimal sequence [14]). However, it can be strongly stimulated by transfection of wild-type (WT) IRF- 3, thus allowing the phenotype of each mutant to be clearly assessed (Fig. 1B, middle panel). In addition, we also tested these mutants as fusion proteins with the Gal4 DNA-binding domain (amino acids 1–147) by doing cotransfections with a reporter, G5E1bCAT, where the chloramphenicol acetyl transferase gene is driven by the E1b TATA box and five copies of a Gal4 binding site (Fig. 1B, right panel). This latter approach minimized the interference due to endogenous IRF proteins associating with the ectopically expressed IRF-3 mutants. Phosphorylation of two distinct groups of Ser/Thr residues is required for virus activation of IRF-3 We mutated the indicated Ser/Thr residues to either alanine (Ala) or glutamic acid (Glu), as shown in Fig. 1B, left panel; in many cases Glu can functionally substitute for phospho-Ser/Thr residues. Mutation of the first set of SerresiduestoeitherAla(IRF-3A2)orGlu(IRF-3E2) drastically reduced the ability of IRF-3 to stimulate P31 ·2 CAT activity in virus-infected cells (middle panel). By contrast, the same constructs behaved differently as Gal4 fusions (right panel). While Gal4-IRF-3A2 showed low, uninducible activity, Gal4-IRF-3E2 displayed signifi- cant activity that was further stimulated by virus infection. Substitution of the second set of Ser/Thr residues had a different impact on IRF-3 activity: IRF-3A5 behaved as WT, except for a slight increase in basal activity; however, the activity of Gal4-IRF-3A5 was significantly stronger than that of the WT construct, but importantly, was still virus-inducible. IRF-3E5 displayed strong basal activity, which was further stimulated by virus infection. Similarly, Gal4-IRF-3E5 had very strong basal activity and virus infection stimulated it further. Simultaneous mutation of both sets of Ser/Thr residues led to IRF-3 mutants (A7, A2E5, E2A5 and E7) that were only marginally inducible upon virus infection by themselves, and not inducible at all as Gal4 fusions. In conclusion, the results that mutation of either set of Ser/Thr residues alone is accompanied by a virus-dependent increase in activity, and that mutation of both sets is not, strongly suggest that residues within both sets of Ser/Thr residues are phos- phorylated in response to virus infection. IRF-3 activates transcription through CBP recruitment We next examined the transcriptional activity of IRF-3 in S2 cells, a Drosophila melanogaster cell line. These cells were chosen because they do not have any apparent IRF homolog, and are therefore unlikely to contain a kinase activity that can specifically phosphorylate the virus- dependent regulatory domain of IRF-3. In these experi- ments, we used a reporter driven by the ISRE of the ISG15 gene, which is a high affinity binding site for IRF-3. We cotransfected pPac plasmids expressing either wild-type or mutant IRF-3 along with ISRE ·3 CAT and in the presence or absence of murine (m)CBP (Fig. 2A). In the presence of mCBP, WT IRF-3 was transcriptionally inactive, but substitution of either or both sets of Ser/Thr residues with Glu(IRF-3E2,E5andE7)ledtosubstantialactivationof the reporter. IRF-3E7 was a much more potent activator than either IRF-3E2 or IRF-3E5. Immunoblot analysis of the IRF-3 mutants indicated that IRF-3E5 and E7 levels were similar, and significantly lower than that of IRF-3 WT or IRF-3E2. Remarkably, we found that none of the IRF-3 constructs activated the reporter in the absence of mCBP, suggesting that IRF-3 by itself is devoid of intrinsic transcriptional activity. Fig. 2. The transcriptional activity of IRF-3 depends on synergy between two sets of residues and is entirely dependent on mammalian CBP. (A) Transcriptional activity in S2 cells of transfected IRF-3 point mutants (0.5 lg) in the presence or absence of cotransfected CBP (1.5 lg) on the ISRE ·3 CAT reporter (top panel) and the expressed proteins were detected by Western blotting (bottom panel; n-sp. stands for nonspe- cific). (B) Transcriptional activity in S2 cells of transfected Gal4-IRF-3 point mutants (1 lg) in the presence or absence of cotransfected CBP (1.5 lg) on the G5E1bCAT reporter (top panel) and the expressed proteins were detected by Western blotting (bottom panel, the top band in each lane corresponds to full-length fusion protein and the same pattern was observed using anti-Gal4 Ig instead of SL12). 6144 H. Yang et al. (Eur. J. Biochem. 269) Ó FEBS 2002 To exclude the influence of possible differences in DNA- binding activity on transcriptional potency, we also tested the same mutants as Gal4 fusions, using G5E1bCAT as reporter plasmid (Fig. 2B). We observed again that none of the Gal4-IRF-3 constructs activated the reporter in the absence of mCBP. Gal4-IRF-3E2 was approximately as strong an activator as Gal4-IRF-3E5 when differences in expression levels were taken into account. IRF-3E7 was the most potent activator of all, either by itself or as a Gal4 fusion, and probably bound the ISRE more effectively than IRF-3E5 in S2 cells as the difference in transcriptional potency between the two mutants was lower when fused to Gal4. Taken together, the results in mammalian and insect cells strongly suggest that residues within both sets must be modified for maximal activation of IRF-3. However, it is unclear why the transcriptional activity of IRF-3E5 was stronger than that of IRF-3E7 in mammalian cells. To address this question, we investigated the dimerization and DNA-binding activities of mutant and WT IRF-3 proteins. Dimerization and DNA-binding activity of IRF-3 mutants Dimerization was assayed by native PAGE in the presence of deoxycholate [22], and DNA-binding was assayed by EMSA using the IFN- and virus-inducible ISRE of the ISG15 gene. We used 293T cells for the following experi- ments because the suggestion that the second set but not the first set of residues was phosphorylated in response to virus infection was based on work using 293T cells [9,19]. First, extracts from transfected 293T cells were assayed by EMSA (Fig. 3A, top panel). Transient expression of WT IRF-3 in 293T cells led to the detection of two new complexes as compared to extracts from cells transfected with empty vector (compare lanes 1 and 2 with 3 and 4). The faster migrating complex corresponds to IRF-3 while the slower complex, with a mobility very similar to that of virus activated factor, corresponds to IRF-3 associated with the p300/CAT coactivators (as determined by supershift experiments [9,11,13–16], data not shown). Both complexes became more intense in the presence of virus infection (compare lanes 3 and 4). The presence of these complexes correlated with the amount of dimeric IRF-3 detected by native PAGE (Fig. 3B, lanes 3 and 4), where untransfected HEC-1B cells extracts were used as a reference to show the change of IRF- 3 mobility that corresponds to a monomer (lane 9, uninfected) to dimer transition (lane 10, virus-infected). Thus, in uninfected but transfected cells, IRF-3 dimerizes and forms the transcriptionally competent complex with p300/CBP (lane 3). By contrast, when cells are not transfected, virus activated factor is only detected in virus- infected cells [14]. Given that DNA transfection alone is sufficient to activate some endogenous IFN-stimulated genes, including ISG15 [25], these results suggest that the formation of the activated IRF-3/coactivator complex in uninfected cells (lane 3) is an artefact of transfection. When a plasmid directing the expression of IRF-3E5 was transfected in 293T cells, both IRF-3 complexes could be detected in control or virus-infected cells by EMSA and virus infection had little effect on the abundance of either complex (compare lanes 5 and 6 in Fig. 3A). However, when IRF-3E7 was expressed in 293T cells, only the IRF- 3/coactivator complex could be detected, and at a level much lower than what was observed for IRF-3E5, despite their similar expression levels (compare lanes 5 and 6 with 7 and 8, bottom panel). In these extracts, approximately half IRF-3E5 was dimeric whether cells were infected or not, while IRF-3E7 was predominantly monomeric (Fig. 3B). Thus, IRF-3E5 expressed by transient transfec- tion dimerized much more efficiently and had a much greater affinity for the ISRE than IRF-3E7 (Fig. 3), consistent with its stronger transcriptional activity (Fig. 1B). Next we investigated the properties of IRF-3 produced by in vitro translation in wheat germ extracts (Fig. 3C). The WT and the E2, E5 and E7 mutant proteins were tested for their ability to dimerize and to bind to the ISRE. Native PAGE indicated that the in vitro produced mutant and WT IRF-3 existed predominantly in the monomeric form (Fig. 3C, top panel) and no specific DNA-binding was detectable under our standard EMSA conditions (data not shown). Thus, the ability of IRF-3E5 to dimerize and bind DNA were significantly different depending on whether it was produced in vitro or in vivo in mammalian cells. While IRF-3E5 affinity for the ISRE was approximately an order of magnitude stronger than that of IRF-3E7 when these proteins were expressed in mammalian cells, the two proteins did not display any significant affinity for the ISRE when produced in vitro. Similarly, native PAGE analysis revealed that half the IRF-3E5 produced in vivo was dimeric, while IRF-3E5 produced in vitro and IRF-3E7, regardless of its source, were predominantly monomeric. Taken together, these results demonstrate that an additional modification of IRF-3E5 took place in vivo in mammalian cells that increased dimerization and DNA-binding activity, presumably phosphorylation of Ser385 and/or Ser386. Fig. 3. IRF-3E5 is further modified upon transfection in mammalian cells. (A) Extracts from 293T cells transfected with empty vector or vector expressing IRF-3 WT or mutants as indicated (lanes 1–8) were submitted to EMSA using the ISG15 ISRE (5 lg extract, top panel) or to immunoblot (IB) analysis using anti-(IRF-3) Ig (SL12) after SDS/ PAGE (10 lg extract, bottom panel); C: control, uninfected cells, V: virus-infected cells. (B) The same extracts as in A (lanes 1–8) and extracts (10 lg) from uninfected (lane 9) or virus-infected (lane 10) HEC-1B cells, were analyzed by native-PAGE and immunoblot ana- lysis using SL12. (C) Proteins were produced by in vitro transcription/ translation using wheat germ extracts and cDNA encoding wild-type and the indicated IRF-3 mutants; control protein (ctrl) is luciferase; translated proteins were detected by immunoblotting using SL12 after native-PAGE (top panel) and SDS/PAGE (bottom panel). Ó FEBS 2002 Mechanism of IRF-3 virus-dependent activation (Eur. J. Biochem. 269) 6145 On the basis of these data, we conclude: (a) maximal dimerization, DNA-binding and transcriptional activity of IRF-3 required modifications within both sets of Ser/Thr residues in the C-terminal virus regulatory domain and (b) IRF-3 has no intrinsic transcriptional activity and depends on its ability to associate with CBP to activate transcription. IRF-3 interacts with multiple domains of the p300/CBP coactivators As shown above, IRF-3 is functionally entirely dependent on p300/CBP for transcriptional activation. We examined the interaction between the coactivators and IRF-3 pro- duced in vivo or in vitro by pull-down/immunoprecipitation assays (Materials and methods). As shown in Fig. 4B, metabolically 32 P-labeled IRF-3 associates specifically with the C-terminal 550 amino acids of mCBP in a virus- dependent manner. Thus, endogenous virus-activated IRF- 3 can interact with the C-terminal domain of CBP. We used proteins produced by in vitro translation in pull-down experiments with immobilized GST fusions for a finer mapping (a representative experiment is shown in Fig. 4C, and binding values referred to in the text below correspond to the average of at least three independent experiments). IRF-3WT bound weakly (1–2% of input) to the N- and C-terminal regions of CBP, and binding to the correspond- ing regions of p300 was even weaker ( 0.5–1% of input). By contrast, binding of either IRF-3E5 or E7 was much stronger than WT: CBP-N,  20%, CBP-C,  24 and 36% of input for IRF-3E5 and E7, respectively. Binding to the corresponding regions of p300 was approximately 2–3 times weaker. When binding of IRF-3 to smaller domains of the N- and C-terminal regions of p300/CBP was examined, most of the activity was found to reside in the N2 and C2 segments. Thus, substitution of Ser/Thr residues with Glu in the virus- regulated domain of IRF-3 led to a strong increase in its affinity for the N2 and C2 regions of the p300 and CBP coactivators. These substitutions partially mimicked the virus-dependent phosphorylation of IRF-3 and allowed us to recapitulate in vitro the association between IRF-3 and p300/CBP that takes place when IRF-3 is activated by virus in vivo (Fig. 4B). The fact that the interactions between IRF-3 and GST- CBP-N were not detected using virus-activated proteins probably reflects the presence of the detergent used to disrupt the interaction between IRF-3/7 and endogenous p300/CBP (Material and methods). Two distinct regions of IRF-3 are required for interaction with coactivators We next examined which regions of IRF-3 are required for interaction with p300 and CBP. IRF-3WT and mutant derivatives were produced by in vitro translation in rabbit reticulocyte lysates and assayed for their ability to bind to the ISRE in the presence or absence of GST-CBP-N or GST-CBP-C2 by EMSA (Fig. 5A). In the absence of GST- CBP fusions, binding of IRF-3 to the ISRE was undetect- able for the WT protein and very weak for IRF-3E7 and the C-terminal truncations 1–409, 1–388 and 1–370 (lanes 4, 7, 10, 13 and 16). Truncations of IRF-3 to amino acids 328, 264 or 241 resulted in much stronger binding to the ISRE (lanes 19, 22 and 25). In the presence of GST-CBP-N a doublet of supershifted bands migrating slowly were detected for all IRF-3 constructs tested. By contrast, in the presence of GST-CBP-C2, the doublet of supershifted bands was detected for only some of the constructs: IRF- 3WT, IRF-3E7 and the truncations to amino acids 409 and 388 (lanes 6, 9, 12 and 15). Truncations to amino acid 370 and shorter did not produce a supershift in the presence of GST-CBP-C2 (lanes 18, 21, 24 and 27). These interactions were also probed in the absence of DNA using pull-down assays (Fig. 5B). Binding of IRF-3 WT to the N- and C-regions of p300 and CBP was weak, while substitution of both sets of Ser/Thr residues with Glu led to much stronger binding for IRF-3E7, in agreement withthedatashowninFig.4C.IRF-3 1)409 , a construct that displays constitutive transcriptional activity in mammalian cells (data not shown), also led to stronger binding to the GST-CBP fusions. IRF-3 truncated to amino acid 388, i.e. between the first and second set of Ser/Thr residues, bound effectively to GST-CBP fusions at a level very close to that observed for IRF-3 1)409 . However, IRF-3 further truncated to amino acids 370 bound poorly to GST-p300C, -CBP-C or -CBP-C2, while binding to GST-p300N, -CBP-N or -CBP-N2 was very similar to that of other IRF-3 Fig. 4. IRF-3 interacts with multiple domains of p300/CBP. (A) Pri- mary structure of mCBP: the position of functional domains is indi- cated, and regions fused to GST protein and used in this study are mapped below. (B) Extracts from control [C] or virus-infected [V] HEC-1B cells labeled in vivo with [ 32 P]orthophosphate were treated with deoxycholate/NP-40 to dissociate IRF proteins from p300/CBP, the detergent concentration was decreased by dilution, and the diluted proteins were incubated with the indicated GST fusions immobilized on glutathione sepharose. Proteins retained on the GST fusions were eluted and immunoprecipitated with anti-(IRF-3) Ig (SL12). Immu- noprecipitated proteins were analyzed by SDS/PAGE and autoradi- ography. (C) 35 S-labeled IRF-3 WT, E5 and E7 were produced by in vitro transcription/translation using rabbit reticulocyte lysates and incubated with the indicated GST fusions of murine CBP and human p300 immobilized on glutathione sepharose for pull-down experi- ments. Proteins retained on the GST fusions were analyzed by SDS/ PAGE and autoradiography. 20% of IRF-3 protein input is shown on the right. Binding to p300-C2 was reproducibly stronger than binding to p300-C. This could be due to poor accessibility to the C2 domain in p300-C, or imperfect folding of p300-C. 6146 H. Yang et al. (Eur. J. Biochem. 269) Ó FEBS 2002 truncations. Taken together, these results indicated that the C-terminal region of IRF-3 including the second set of Ser/ Thr residues is dispensable for binding to the C2 region of the p300 and CBP coactivators, and that the C-terminal end-point of this interaction domain is located between amino acids 370 and 388. By contrast, binding to the N2 region of CBP could be achieved with only the first 241 amino acids of the protein. Mapping IRF-3 dimerization domain IRF-3 truncations’ ability to bind DNA varied considerably depending on the end-point of each truncation. Thus, full- length protein produced by in vitro translation does not bind DNA, while truncation at amino acids 328 resulted in strong binding (Fig. 5A [14]). We generated a series of truncations at  20 amino acid intervals to map the domains of IRF-3 required for DNA binding and dimeri- zation more precisely (Fig. 5C,D). When analyzed by deoxycholate-PAGE and immunoblotting, IRF-3 WT pro- duced in vitro was predominantly in a faster migrating form, as observed in Fig. 3C. Progressive truncations from the C-terminus led to the detection of slower migrating forms (Fig. 5C, top panel). In the case of virus-activated IRF-3 (Fig. 3B), the slower migration on deoxycholate-PAGE could be due to dimerization of the protein, or simply to the phosphorylation of its C-terminus. However, the latter is unlikely as the proteins are produced in vitro and because the target residues are absent from IRF-3 1)368 and shorter truncations. Rather, the slower migrating forms most likely corresponded to dimers and higher order oligomers. The same truncations were assayed for their ability to bind the ISRE by EMSA, and the amount of DNA-binding, normalized to the amount of protein, is charted in Fig. 5D, along with the ratio of dimeric to monomeric forms. Truncation to amino acids 409 or 388 resulted in a small increase in the proportion of the dimeric form and in low levels of detectable DNA-binding activity, as compared to full-length IRF-3. Further truncation resulted in a much higher proportion of the dimeric form and in higher levels of binding to the ISRE up to amino acids 328–308. IRF-3 1)288 (Fig. 5D) and shorter forms (Fig. 5A and data not shown) displayed reduced DNA-binding and dimerization. Thus, there was a strong correlation between the ability of IRF-3 to dimerize and its ability to bind DNA. These data, together with previous results [14,19], suggest that progres- sive truncations from the C-terminus of IRF-3 removed a domain that prevented dimerization, and the ability to bind DNA that accompanied it. Further truncations eventually affected the dimerization domain whose C-terminal end- point is located between amino acids 308 and 288 (Fig. 5). Each of IRF)3¢s multiple interactions with coactivators is essential for activity We have shown above that IRF-3 transcriptional activity was entirely dependent on its ability to associate with mammalian p300/CBP (Fig. 2), and that IRF-3 physically interacted with distinct regions of these coactivators (Figs 4 and 5). The importance of each of these contacts for IRF-3¢s ability to activate transcription was tested in S2 cells. We first investigated which IRF-3 truncations would be active in insect cells (Fig. 6A). Interestingly, both IRF-3 1)409 and IRF-3 1)388 significantly activated transcription of the ISRE ·3 CAT reporter in the presence of mammalian coac- tivators, and these truncations interacted with both the N- and C-terminal regions of p300/CBP (Fig. 5A,B). By contrast, IRF-3 1)370 and IRF-3 1)328 interacted significantly only with the N-terminus of p300/CBP and displayed no detectable transcriptional activity. Thus, even though IRF- 3 1)328 dimerized and bound the ISRE much more efficiently than IRF-3 1)409 or IRF-3 1)388 , it failed to activate tran- scription in the absence of contact with the C-terminal part of CBP. Next, we investigated the ability of N- and C-terminal fragments of CBP to interfere with the ability of IRF-3E7 and full-length CBP to activate transcription (Fig. 6B). Expression of GST-CBP 1)1100 and GST-CBP 1892)2441 Fig. 5. Two domains of IRF-3 are involved in interactions with coacti- vators. (A) Proteins were produced by in vitro transcription/translation using rabbit reticulocyte lysates and cDNAs encoding WT and the indicated IRF-3 mutants; control protein (ctrl) is luciferase; translated proteins were analyzed in the presence of GST, GST-CBP-N and GST-CBP-C2 by EMSA using the ISG15 ISRE as a probe (left panel); translated proteins were detected by immunoblotting (right panel) after SDS/PAGE. (B) 35 S-labeled IRF-3 WT, E7, 1–409, 1–388 and 1–370 were produced by in vitro transcription/translation and incubated with the indicated GST fusions immobilized on glutathione sepharose for pull-down experiments. Proteins retained on the GST fusions were analyzed by SDS/PAGE and autoradiography. Twenty-five per cent of IRF-3 proteins input is shown on the right. (C) Proteins were produced by in vitro transcription/translation using rabbit reticulocyte lysates and cDNAs encoding WT and the indicated IRF-3 truncations; con- trol protein (ctrl) is luciferase; translated proteins were analyzed by deoxycholate-PAGE (top panel) or SDS/PAGE (bottom panel) and immunoblotting with SL12; the wavy pattern of migration for the various IRF-3 truncations on deoxycholate-PAGE was highly repro- ducible. (D) Proteins used in C were analyzed by EMSA as in A, and the amount of protein bound to the ISRE, expressed in arbitrary units (open triangles), was plotted along the ratio of dimeric to monomeric IRF-3 (filled squares) determined from quantification of C. Ó FEBS 2002 Mechanism of IRF-3 virus-dependent activation (Eur. J. Biochem. 269) 6147 reduced transcriptional activation by up to 95 and 70%, respectively, suggesting that these fusion proteins effectively interacted with IRF-3E7 in transfected insect cells. In the absence of full-length mCBP, coexpression of IRF-3E7 with GST-CBP 1)1100 , GST-CBP 1892)2441 or their combination did not result in any activation of the ISRE · 3CAT reporter (Fig. 6B). The inability of GST-CBP 1)1100 and GST-CBP 1892)2441 , alone or in combination, to activate transcription together with IRF-3E7 was not due to a lack of transcription potential of these CBP fragments. Indeed, expression of various Gal4-mCBP fusion constructs in insect cells revealed that, as observed in mammalian cells [26], both the N- and C-termini of mCBP displayed intrinsic transcriptional activity (Fig. 6C). Furthermore, coexpres- sion of GST-CBP 1)1100 or GST-CBP 1892)2441 had minimal effects on the transcriptional activity of Gal4-mCBP 1)2441 (Fig. 6D), demonstrating that the inhibitory effect on the activity of the IRF-3/CBP complex was not due to interference with CBP transcriptional activity but to inter- ference with the interaction between IRF-3 and CBP. Taken together, these results demonstrate that the transcriptional activity of IRF-3 is dependent on simultaneous contact with both the N- and C-termini of CBP, and on the physical integrity of CBP. We also tested the ability of IRF-3E7 and human p300 or mutant derivatives [27] to activate transcription from the ISRE ·3 CAT reporter in S2 cells (Fig. 6E,F). The level of activation achieved by IRF-3 and p300 WT was approxi- mately twofold lower than that reached by the IRF-3/CBP combination. Deletion of the p300 Bromo domain (DBromo, D amino acids 1071–1241) resulted in a significant reduction of p300s ability to activate transcription in combination with IRF-3. The other deletions tested, p300DNR (D amino acids 3–173), p300DE1a (D amino acids 1739–1871) and p300DSRC (D amino acids 2042– 2157) all completely failed to activate transcription in the presence of IRF-3. The inability of p300DSRC to activate transcription with IRF-3 was expected, as this deletion removes one of the two major interaction regions with IRF- 3. The failure of p300DNR and IRF-3 to activate transcription might similarly reflect a decreased affinity of p300 for IRF-3 upon removal of part of its N–terminal interaction domain. The basis for p300DE1a lack of activity in this assay remains to be determined. However, this result suggests that the E1a region of p300, which is known to interact with general transcription factors such as TFIIB, p/ CAF and RNA polymerase II [28–31], also plays an essential role in IRF-3 transcriptional activity. DISCUSSION Previous studies have identified IRF-3 and IRF-7 as essential mediators of the transcriptional response in virus- infected vertebrate cells [9–14,16,32–34]. Indeed, each protein becomes hyperphosphorylated following virus infection, dimerizes, accumulates in the nucleus and activates transcription of a specific set of genes. Moreover, in vivo both proteins are found to be physically associated with the promoter of the IFI-56K and IFN-b genes, in a virus-dependent manner [14]. Finally, gene targeting experiments further demonstrated that IRF-3 and IRF-7 play essential yet distinct roles in the response to virus infection [33,35]. However, detailed investigations of the mechanism by which IRF-3 and IRF-7 activate transcrip- tion have been hampered by a number of limitations inherent to the experimental systems used. These include: (a) the presence of endogenous IRFs, which can associate with transfected IRFs; (b) the presence of the endogenous p300/CBP coactivators and the unavailability of animals or cell lines null for them due to embryonic and cellular lethality; (c) the existence of a feedback loop involving virus-induced IFN that leads to the formation of ISGF3 and the induction of both IRF-1 and IRF-7, all transcription factors that can activate the reporters used to monitor the transcriptional activity of IRF-3; (d) the ability of DNA transfection per se to undesirably stimu- late the virus-activated signal transduction pathway to some extent and (e) the presence of viral gene products in certain cell lines that can potentially interfere with p300/ CBP function (e.g. E1a and SV40 large T in 293T cells; Fig. 6. All interactions between IRF-3 and coactivators are essential for transcriptional activity. (A) Transcriptional activity in S2 cells of transfected IRF-3 deletion mutants (0.5 lg) on the ISRE ·3 CAT re- porter in the presence of cotransfected p300/CBP (1.5 lg). (B) S2 cells were transfected with IRF-3E7 (0.5 lg), mCBP (0.5 lg) and the in- dicated GST-CBP fusions (0.5 and 2 lg in the presence of CBP, 2 lgin its absence) together with the ISRE ·3 CAT reporter. (C) Transcrip- tional activity in S2 cells of the indicated Gal4-mCBP fusions (2 lg) on the G5E1bCAT reporter. (D) S2 cells were transfected with Gal4- mCBP 1)2441 (0.5 lg), the indicated GST-CBP fusions (0.5 and 2 lg) together with the G5E1bCAT reporter. (E) Transcriptional activity in S2 cells of transfected IRF-3E7 (0.5 lg) on the ISRE ·3 CAT reporter in the presence of cotransfected p300 (1.5 lg), WT or the indicated mu- tants, schematically represented in (F), DNR (D3–173), DBromo (D1071–1241), DE1a (D1739–1871), DSRC (D2042–2157) [27]. 6148 H. Yang et al. (Eur. J. Biochem. 269) Ó FEBS 2002 SV40 large T in COS cells). In this paper, we have examined the molecular events by which IRF-3 activates transcription in response to virus infection using a combination of approaches that circumvent the limitations discussed above, allowing us to reconcile previously conflicting interpretations and to further define the mechanism by which IRF-3 activates transcription. Modification of both sets of Ser/Thr residues is essential for full activation of IRF-3 In SAN cells, the transcriptional activity of essentially all Gal4-IRF-3 constructs where only one set was mutated was still virus-inducible, while the activity of all constructs where both sets were mutated was not (Fig. 1B), suggesting that some residues, within each set, are phosphorylated upon infection. This conclusion is strongly supported by the observation that IRF-3E7 is much more active than either IRF-3E2 or IRF-3E5 in insect cells, where the absence of additional post-translational modifications of IRF-3E5 uncovers the synergy between the two sets of Ser/Thr residues in the activation of IRF-3 (Fig. 2). In contrast to SAN cells, constructs like Gal4-IRF-3E5 are constitutively active in 293T cells, with no further stimulation by virus ([19], data not shown). Together with the observation that IRF-3E5 is a stronger activator than IRF-3E7inmammaliancells,theseresultsledtothe conclusion that the first set of residues is not phosphory- lated upon virus infection, but might play a regulatory role in the phosphorylation events at the C-terminal end of IRF-3, e.g. they could be part of the surface of the protein recognized by the virus-activated kinase that would phosphorylate the downstream Ser/Thr residues [19]. However, IRF-3E5 has no constitutive activity in L929 cells [18]. Thus, the transcriptional potential of IRF- 3E5 and its virus-dependence are cell type specific. What accounts for these differences? One possibility is that IRF- 3E5 could be additionally modified when transfected in mammalian cells. 293T cells can be transfected with very high efficiencies, and a high transfection efficiency in turn would result in high levels of dsRNA production by symmetric transcription of the transfected plasmids (limi- tation #4). Accordingly, subsequent virus infection would not result in any further increase in transcriptional activity. Alternatively, it is possible that substitution to E5 primes IRF-3 for its phosphorylation either by the genuine virus-activated IRF-3 kinase or by another endogenous kinase. The possibility that IRF-3E5 is further modified upon transfection into mammalian cells has considerable support from the DNA binding and dime- rization experiments (Figs 3 and 5). That is, IRF-3E5 dimerized and bound the ISRE much more effectively than IRF-3E7 when these proteins were produced in transfected 293T cells, a difference that was absent in wheat germ extracts (or in insect cells). Thus, IRF-3E5 can be additionally modified in some transfected mam- malian cells, and this modification is presumably phos- phorylation of Ser385/386 as mutation of these residues to either Ala or Glu led to much weaker DNA binding. Additional evidence that the first set of Ser residues is phosphorylated in response to virus infection comes from studies where IRF-3A5 can be in vitro phosphorylated, but only using extracts from virus-infected cells [22]. Mechanism of IRF-3 activation IRF-3 exists in primarily two forms, one of which is monomeric, exhibits weak affinity for DNA or p300/CBP, and shuttles between the cytoplasm and the nucleus, with a dominance of export over import. Another form of IRF-3 is dimeric, binds efficiently to DNA, interacts strongly with the p300/CBP coactivators and resides in the nucleus. The equilibrium between the monomeric and dimeric forms of IRF-3 is affected by phosphorylation of the Ser/Thr residues in its C-terminal regulatory domain. When these residues are unphosphorylated, the monomeric form of IRF-3 is predominant, while phosphorylation shifts the equilibrium towards the dimeric form. When IRF-3 is in its monomeric form, there is an intramolecular interaction involving the C-terminal domain and a region C-terminal of the DNA-binding domain (amino acids 98–240). Lin et al. [19] mapped the minimal C-terminal domain to amino acids 380–427, and our results suggest that maximal interaction involves amino acids 328– 427). The region (amino acids 98–240) with which the C-terminal domain interacts is part of IRF-3 dimerization domain, whose C-terminal end point we mapped between amino acids 288 and 308. Interference with this intramolecular interaction leads to the activation of IRF-3. In vitro, removal of only the C-terminal 17 residues is sufficient to lead to low levels of dimerization, ISRE–binding and interaction with coactiva- tors (Fig. 5), resulting in transcriptional activity in insect (Fig. 6) or mammalian cells (data not shown). In vivo, this intramolecular interaction is naturally disrupted when IRF-3 becomes phosphorylated in virus- infected cells. This conformational change involves the loss and gain of molecular interactions, and the two sets of phosphorylated residues could play distinct roles in this process. IRF-3 dimerizes and binds to DNA much more efficiently when the first set is phosphorylated than when it is not or when it is substituted with Ala or Glu residues (Figs 1 and 3). These results thus suggest the first set of Ser residues are involved in a new intra- or intermolecular interaction where phosphorylated Ser385/386 interact with another domain of IRF-3 either on the same molecule or on the dimerization partner, and this gain of interaction can only be inefficiently mimicked by glutamic (or aspartic) acid substitution. By contrast, phosphorylation of the second set of Ser/Thr residues seems to be primarily involved in the loss of the intramolecular interaction that keeps IRF-3 in the inactive form. Indeed, ectopic expression of IRF-3 WT, IRF-3A5 and IRF-3E5 lead to very similar levels of transcriptional activation from the P31 ·2 CAT reporter in virus-infected SAN cells (Fig. 1B). Therefore, the second set participates minimally in intra- or intermolecular interac- tionswhenIRF-3isadimerasAla,Gluorphospho-Ser/ Thr residues within this set all displayed the same phenotype in these assays. However, unphosphorylated Ser/Thr resi- dues within the second set must participate in direct contacts with the amino acids 98–240 domain in the monomeric form, as this intramolecular interaction can be disrupted by substitution to either Ala or Glu. It is important to note that we do not claim that all residues become phosphorylated in either set. Rather, our experiments only suggest that residues within each set are modified upon virus infection and that there is a functional synergy between phosphorylated Ó FEBS 2002 Mechanism of IRF-3 virus-dependent activation (Eur. J. Biochem. 269) 6149 residues present in these sets. Additional experiments are required to determine exactly which residues within each set become phosphorylated in infected cells. Interaction with a coactivator is essential for IRF-3 transcriptional activity Unprecedently, the transcriptional activity of IRF-3 is entirely dependent on its interaction with a mammalian coactivator as demonstrated by transfection experiments in insect cells (Figs 2 and 6), and is consistent with the observation that E1a can strongly interfere with IRF-3 activity in mammalian cells [36]. Intriguingly, while the yeast genome contains no CBP homolog, a Gal4-IRF-3 fusion is transcriptionally active in yeast cells ([9] and our unpub- lished results). Even full-length IRF-3 fused to Gal4 is active in those cells, unlike what we observed in insect cells in the absence of mCBP (Fig. 3B). Mapping of the IRF-3 activation domain in yeast identified residues 134–394 as the minimum region required for transcriptional activity, an unusually large activation domain as noted by the authors [19]. This domain contains amino acids 139–386, which is the minimal domain required for interaction with the IBiD in CBP [37] (Figs 4 and 5). Because Gal4 binds DNA as a dimer, the dimeric conformation of IRF-3 is presumably favored in these experiments, thus exposing a region of the protein that is not accessible under physiological conditions, either because IRF-3 is in its monomeric conformation or because it interacts with p300/CBP. It is possible that the transcription potential of IRF-3 in yeast cells is due to a spurious interaction between this region of IRF-3 and a component of the yeast transcription machinery. Virus-dependent phosphorylation of IRF-3 leads to strong association with p300 and CBP [12,14–16,18,38], and this association involves multiple interactions (Figs 4– 6). There is an interaction between the N-terminal half of IRF-3 (amino acids 1–241) and N-terminal fragments of p300 and CBP. This was further mapped to CBP-N2 (amino acids 267–462 of CBP), which contains the CH1/ TAZ1 domain. This region of the coactivator is known to interact with a large number of transcription factors, including RelA, STAT-2 and p53. There is another interaction between a central region of IRF-3 (amino acids 139–386) and the C-terminal part of p300 and CBP, more specifically the C2 region that contains the recently described IBiD domain, which is known to interact with TIF-2, Ets2 and E1a ([37] and references therein). The first set of Ser residues may not be directly involved in binding of CBP/p300, as changing them both to Ala or Glu had little effect on the association of IRF-3 1)388 with GST-CBP-C2 in vitro (data not shown). However, a peptide extending from amino acids 375–427 could compete with the interac- tion between GSTDp300 (amino acids 1752–2221) and virus-activated IRF-3 only when the peptide was phosphor- ylated at position 385 and 386. Based on this latter result, it was proposed that when Ser385/386 are phosphorylated, these residues make direct contact with p300 [18]. It is possible though it appears not very likely that phospho- Ser385/386 could make simultaneous contacts with another region of IRF-3 and with p300/CBP. As phosphorylation of Ser385/386 seems to be required for a new interaction that promotes both dimerization and DNA binding, even in the absence of p300 or CBP, an alternative or complementary explanation for the effect of the phosphorylated peptide on the interaction between IRF-3 and p300 would be that this latter interaction is dependent on IRF-3 adopting the dimeric conformation. In this scenario, the peptide would be competing for the intra- or intermolecular interaction involving phospho-Ser385/386 and another domain of IRF-3, preventing dimerization and hence interaction with p300/CBP. CONCLUSIONS In summary, IRF-3 is phosphorylated on two sets of Ser/ Thr residues within its C-terminus upon virus infection. Phosphorylation of both sets is functionally important for full dimerization, DNA-binding, p300/CBP interaction and transcriptional activity, and each set might play distinct roles in the conformational switch that accounts for IRF-3 activation. Activated IRF-3, in turn, entirely depends on simulta- neous interactions with multiple domains of the p300/CBP coactivators to stimulate transcription. Simply recruiting the coactivators’ intrinsic transcription potential to IRF-3 is not sufficient, as shown by the failure of GST-CBP 1)1100 or GST-CBP 1892)2441 alone or in combination to activate transcription together with IRF-3E7 in insect cells. By contrast, such fragments are sufficient to stimulate the activity of other transcription factors [39]. This failure is not due to an inability of CBP 1-1100 or CBP 1892-2441 to (a) interact with IRF-3E7 in the transfected cells (as they can interfere with transcriptional activation mediated by IRF- 3E7 and full-length mCBP), or to (b) independently activate transcription(astheycandosowhenfusedtotheGal4 DNA-binding domain, Fig. 6). Rather, these results and the effect of p300 deletions underscore how multiple interac- tions between IRF-3 and a mammalian coactivator are indispensable to activate transcription. ACKNOWLEDGEMENTS We would like to thank T. Collins, R. Goodman, W. Lee Krauss and D. Livingston for kindly providing reagents, and Maria Czyzyk- Krzeska and Nelson Horseman for critical reading of the manuscript. This work was supported by a Dean Research Award to M. G. W., and by grant from the National Institutes of Health (AI20642) to Tom Maniatis, Harvard University, during its initial phase. REFERENCES 1. Kimbrell, D.A. & Beutler, B. (2001) The evolution and genetics of innate immunity. Nature Rev. Genet. 2, 256–267. 2. Stewart, W.E. II (1979) The Interferon System. Springer-Verlag, Wien and New York. 3. Mamane, Y., Heylbroeck, C., Genin, P., Algarte, M., Servant, M.J.,LePage,C.,DeLuca,C.,Kwon,H.,Lin,R.&Hiscott,J. (1999) Interferon regulatory factors: the next generation. Gene 237, 1–14. 4. Stark, G.R., Kerr, I.M., Williams, B.R., Silverman, R.H. & Schreiber, R.D. (1998) How cells respond to interferons. Annu. Rev. Biochem. 67, 227–264. 5. Taniguchi, T., Ogasawara, K., Takaoka, A. & Tanaka, N. (2001) IRF family of transcription factors as regulators of host defense. Annu. Rev. Immunol. 19, 623–655. 6. Maniatis, T., Falvo, J.V., Kim, T.H., Kim, T.K., Lin, C.H., Parekh, B.S. & Wathelet, M.G. (1998) Structure and Function of 6150 H. Yang et al. (Eur. J. Biochem. 269) Ó FEBS 2002 the Interferon-b Enhanceosome. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. 7. Navarro, L. & David, M. (1999) p38-dependent activation of interferon regulatory factor 3 by lipopolysaccharide. J. Biol. Chem. 274, 35535–35538. 8. Au, W.C., Moore, P.A., LaFleur, D.W., Tombal, B. & Pitha, P.M. (1998) Characterization of the interferon regulatory factor-7 and its potential role in the transcription activation of interferon A genes. J. Biol. Chem. 273, 29210–29217. 9. Lin, R., Heylbroeck, C., Pitha, P.M. & Hiscott, J. (1998) Virus- dependent phosphorylation of the IRF-3 transcription factor regulates nuclear translocation, transactivation potential, and proteasome-mediated degradation. Mol. Cell Biol. 18, 2986–2996. 10. Marie, I., Durbin, J.E. & Levy, D.E. (1998) Differential viral induction of distinct interferon-alpha genes by positive feedback through interferon regulatory factor-7. EMBO J. 17, 6660–6669. 11. Sato, M., Hata, N., Asagiri, M., Nakaya, T., Taniguchi, T. & Tanaka, N. (1998a) Positive feedback regulation of type I IFN genes by the IFN-inducible transcription factor IRF-7. FEBS Lett. 441, 106–110. 12. Sato, M., Tanaka, N., Hata, N., Oda, E. & Taniguchi, T. (1998b) Involvement of the IRF family transcription factor IRF-3 in virus- induced activation of the IFN-beta gene. FEBS Lett. 425, 112–116. 13. Schafer, S.L., Lin, R., Moore, P.A., Hiscott, J. & Pitha, P.M. (1998) Regulation of type I interferon gene expression by inter- feron regulatory factor-3. J. Biol. Chem. 273, 2714–2720. 14. Wathelet, M.G., Lin, C.H., Parekh, B.S., Ronco, L.V., Howley, P.M. & Maniatis, T. (1998) Virus infection induces the assembly of coordinately activated transcription factors on the IFN-b enhancer in vivo. Mol. Cell. 1, 507–518. 15. Weaver, B.K., Kumar, K.P. & Reich, N.C. (1998) Interferon regulatory factor 3 and CREB-binding protein/p300 are subunits of double-stranded RNA-activated transcription factor DRAF1. Mol. Cell Biol. 18, 1359–1368. 16. Yoneyama, M., Suhara, W., Fukuhara, Y., Fukuda, M., Nishida, E. & Fujita, T. (1998) Direct triggering of the type I interferon system by virus infection: activation of a transcription factor complex containing IRF-3 and CBP/p300. EMBO J. 17, 1087– 1095. 17. Barnes, B.J., Moore, P.A. & Pitha, P.M. (2001) Virus-specific activation of a novel interferon regulatory factor, IRF-5, results in the induction of distinct interferon alpha genes. J. Biol. Chem. 276, 23382–23390. 18. Suhara, W., Yoneyama, M., Iwamura, T., Yoshimura, S., Tam- ura, K., Namiki, H., Aimoto, S. & Fujita, T. (2000) Analyses of virus-induced homomeric and heteromeric protein associations between IRF-3 and coactivator CBP/p300. J. Biochem. (Tokyo) 128, 301–307. 19. Lin, R., Mamane, Y. & Hiscott, J. (1999) Structural and func- tional analysis of interferon regulatory factor 3: localization of the transactivation and autoinhibitory domains. Mol. Cell Biol. 19, 2465–2474. 20. Lin, R., Mamane, Y. & Hiscott, J. (2000) Multiple regulatory domains control IRF-7 activity in response to virus infection. J. Biol. Chem. 275, 34320–34327. 21. Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) Molecular Cloning: a Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. 22. Iwamura,T.,Yoneyama,M.,Yamaguchi,K.,Suhara,W.,Mori, W., Shiota, K., Okabe, Y., Namiki, H. & Fujita, T. (2001) Induction of IRF-3/-7 kinase and NF-kappaB in response to double- stranded RNA and virus infection: common and unique pathways. Genes Cells 6, 375–388. 23. Harlow, E. & Lane, D. (1988) Antibodies: a Laboratory Manual. Cold Spring Harbor Press, Cold Spring Harbor, New York. 24. Gerritsen, M.E., Williams, A.J., Neish, A.S., Moore, S., Shi, Y. & Collins, T. (1997) CREB-binding protein/p300 are transcriptional coactivators of p65. Proc. Natl Acad. Sci. USA 94, 2927–2932. 25. Pine, R., Levy, D.E., Reich, N. & Darnell, J.E. Jr (1988) Tran- scriptional stimulation by CaPO4-DNA precipitates. Nucleic Acids Res. 16, 1371–1378. 26. Swope, D.L., Mueller, C.L. & Chrivia, J.C. (1996) CREB-binding protein activates transcription through multiple domains. J. Biol. Chem. 271, 28138–28145. 27. Kraus, W.L., Manning, E.T. & Kadonaga, J.T. (1999) Biochem- ical analysis of distinct activation functions in p300 that enhance transcription initiation with chromatin templates. Mol. Cell Biol. 19, 8123–8135. 28. Yang, X.J., Ogryzko, V.V., Nishikawa, J., Howard, B.H. & Na- katani, Y. (1996) A p300/CBP-associated factor that competes with the adenoviral oncoprotein E1A. Nature 382, 319–324. 29. Kwok, R.P., Lundblad, J.R., Chrivia, J.C., Richards, J.P., Ba- chinger, H.P., Brennan, R.G., Roberts, S.G., Green, M.R. & Goodman, R.H. (1994) Nuclear protein CBP is a coactivator for the transcription factor CREB. Nature 370, 223–226. 30. Yuan, W., Condorelli, G., Caruso, M., Felsani, A. & Giordano, A. (1996) Human p300 protein is a coactivator for the transcrip- tion factor MyoD. J. Biol. Chem. 271, 9009–9013. 31. Nakajima, T., Uchida, C., Anderson, S.F., Lee, C.G., Hurwitz, J., Parvin, J.D. & Montminy, M. (1997) RNA helicase A med- iates association of CBP with RNA polymerase II. Cell 90, 1107– 1112. 32. Navarro, L., Mowen, K., Rodems, S., Weaver, B., Reich, N., Spector, D. & David, M. (1998) Cytomegalovirus activates inter- feron immediate-early response gene expression and an interferon regulatory factor 3-containing interferon-stimulated response ele- ment-binding complex. Mol. Cell Biol. 18, 3796–3802. 33. Sato, M., Suemori, H., Hata, N., Asagiri, M., Ogasawara, K., Nakao, K., Nakaya, T., Katsuki, M., Noguchi, S., Tanaka, N. & Taniguchi, T. (2000) Distinct and essential roles of transcription factors IRF-3 and IRF-7 in response to viruses for IFN-alpha/ beta gene induction. Immunity 13, 539–548. 34. Weaver, B.K., Ando, O., Kumar, K.P. & Reich, N.C. (2001) Apoptosis is promoted by the dsRNA-activated factor (DRAF1) during viral infection independent of the action of interferon or p53. FASEB J. 15, 501–515. 35. Nakaya, T., Sato, M., Hata, N., Asagiri, M., Suemori, H., Noguchi, S., Tanaka, N. & Taniguchi, T. (2001) Gene induction pathways mediated by distinct IRFs during viral infection. Bio- chem. Biophys. Res. Commun. 283, 1150–1156. 36. Juang, Y., Lowther, W., Kellum, M., Au, W.C., Lin, R., Hiscott, J. & Pitha, P.M. (1998) Primary activation of interferon A and interferon B gene transcription by interferon regulatory factor 3. Proc. Natl Acad. Sci. USA 95, 9837–9842. 37. Lin,C.H.,Hare,B.J.,Wagner,G.,Harrison,S.C.,Maniatis,T.& Fraenkel, E. (2001) A small domain of CBP/p300 binds diverse proteins: solution structure and functional studies. Mol. Cell 8, 581–590. 38. Suhara, W., Yoneyama, M., Kitabayashi, I. & Fujita, T. (2002) Direct involvement of CREB-binding protein/p300 in sequence- specific DNA binding of virus-activated interferon regulatory factor-3 holocomplex. J. Biol. Chem. 277, 22304–22313. 39. Zanger, K., Cohen, L.E., Hashimoto, K., Radovick, S. & Won- disford, F.E. (1999) A novel mechanism for cyclic adenosine 3¢,5¢- monophosphate regulation of gene expression by CREB-binding protein. Mol. Endocrinol. 13, 268–275. Ó FEBS 2002 Mechanism of IRF-3 virus-dependent activation (Eur. J. Biochem. 269) 6151 . Transcriptional activity of interferon regulatory factor (IRF)-3 depends on multiple protein–protein interactions Hongmei Yang 1 , Charles. activity. Fig. 2. The transcriptional activity of IRF-3 depends on synergy between two sets of residues and is entirely dependent on mammalian CBP. (A) Transcriptional